[00:03]
Perplexity just announced something that most people aren't talking about yet. And once you understand what it is, it completely changes how you think about where AI is going. They called it Perplexity Frontier Research. And
[00:17]
they're backing it with over 10,000 GPUs. That's not a small side project. That's a full research program announced on November 16th, 2026 with a specific
[00:29]
goal. Build an AI system that can do research on its own. Find the question. Search thousands of sources. Build a theory. Test it. See what works. Adjust.
[00:41]
Keep going. All without a human holding its hand. And the reason this matters for you, especially if you're building any kind of business with AI automation, is because the research Perplexity is doing right now feeds directly into the
[00:55]
tools millions of people use every single day. This isn't a lab thing that never touches the real world. It already has. So, let's break down what Perplexity actually announced, why it's a bigger deal than the headlines
[01:09]
suggest, and then I'll show you four practical workflows you can run today using Perplexity's current tools, all built around growing an AI automation business. First, some quick context on where Perplexity started. When it
[01:24]
launched, it was basically a smarter version of Google. You typed a question, it searched the web, it gave you a cleaner answer. That was genuinely useful, but the product kept evolving fast. In February 2026, they launched
[01:40]
Perplexity Computer. That was a step change. Instead of one AI model answering questions, Plexity Computer runs 19 different Frontier AI models at once, rooting each part of a task to the model best suited for it. Claude Opus
[01:56]
model best suited for it. Claude Opus 4.6 handles the core reasoning. Gemini runs the deep research layers. Other models handle writing, analysis, coding, and more. It's not one AI doing everything. It's a coordinated team. And
[02:11]
then they moved deep research into computer. This is the feature that takes a complex question, breaks it into multiple subtasks, roots each subtask across more than 20 frontier models, and returns a full research report. The
[02:26]
accuracy numbers jumped hard when this happened. On one benchmark called Humanity's Last Exam, the score went from 36.4% to 50.5%. On a web research benchmark called
[02:40]
On a web research benchmark called Browse Comp, it went from 40.7% to 83.8%. Those are not small improvements. That's a system that is fundamentally better at doing real research. So, the trajectory
[02:54]
looks like this. search engine, then multimodel computer, then now building an autonomous AI research system. Each step makes the previous one look simple. Let me just pause here because there's something directly relevant to this.
[03:09]
Inside the AI profit boardroom, inside the AI profit boardroom, we've built a 30-day road map specifically around using Perplexity for research and AI automation workflows in your business. Day one, you set up perplexity and run
[03:24]
your first real research task, competitor analysis for your niche. Day two, you build a lead research workflow using deep research to find ideal clients with detailed prospect profiles. Day three, you connect perplexity into a
[03:40]
content pipeline, so your research automatically feeds into what you publish. Each day builds one more piece until you have a full AI powered research system working in your business. We run four live coaching
[03:53]
calls every week where members bring their actual Perplexity setups, ask questions live, and go deep on what's working in their specific situation. A lot of members inside are already running Perplexity Computer for client
[04:07]
research, market intelligence, and lead generation right now, and they're sharing results in real time inside the community. If you want to build this step by step, the road map is already in there. link in the description or go to
[04:22]
aiprofitboardroom.com. Now, let's talk about what PerplexityFrontier Research actually is in plain English. The goal is to build what they call an automated research scientist. Think about what a human
[04:36]
researcher does. They don't just answer the questions you give them. They find questions worth asking. They search through existing work to find gaps. They form a theory. They test it. They look at what came back and change their
[04:50]
approach if it didn't work and they keep going until they find something real. Perplexity wants to build AI that runs that entire process on its own. Andrew Gordon Wilson, an NYU professor, was specifically hired to lead this. His
[05:06]
background is in continual learning, which is teaching AI to keep learning from new experiences without forgetting what it already knows. That's one of the core technical challenges in building AI that actually improves over time instead
[05:20]
of staying frozen at whatever it knew during training. They're doing this research across six specific areas. And understanding these six gives you a clear picture of where AI tools are going, including the ones you're using
[05:33]
right now. The first area is multi- aent orchestration. This is getting different AI agents to work as a team. Each agent has a specific job. One searches, one has a specific job. One searches, one analyzes, one challenges the findings,
[05:48]
one writes the output, one checks the work. Right now, most AI tools are solo, one model doing everything. Multi-agent orchestration is what turns AI from a
[06:00]
solo worker into a team. The second is pre-training. This is improving how AI models are built from the ground up. Better foundations mean more capable AI across every task. It's the deep work that makes everything else possible. The
[06:16]
third is continual learning. This is a big one. Right now, most AI models are trained once and then deployed. They don't learn from using them. Continual learning means the AI gets better every time it works on something new without
[06:31]
forgetting what it already knew. Andrew Gordon Wilson is specifically focused here and it's one of the most important pieces of building AI that actually improves from experience. The fourth is reinforcement learning. The AI tries
[06:47]
something, gets feedback, learns what worked and what didn't, tries again. This is how you build AI that improves through trial and error. Same way humans learn. It's already used in many AI systems today, but perplexity is pushing
[07:02]
systems today, but perplexity is pushing it further. The fifth is systems. This is the infrastructure layer. The nuts and bolts that let the AI run fast, efficiently, and at scale. Perplexity actually published research on this
[07:15]
already. They built custom networking systems between GPUs that hit 97.1% bandwidth efficiency. That is what allows Perplexity Computer to run 19
[07:27]
models in a single workflow without grinding to a halt. and they open sourced a library called fabric lib from this work so the broader AI community can use it. The sixth is frontier evaluations. This is about testing
[07:42]
whether AI can actually reason and discover things, not just memorize something that isn't in its training data? Can it reason through a genuinely new problem? This is how you know you're building something that's actually
[07:57]
getting smarter rather than just bigger. six areas all interconnected, all feeding into products that you can use right now and products that are going to be significantly more powerful in the next 12 months. There's a window right
[08:11]
now where most business owners haven't figured out how to use AI research tools properly. They know the name perplexity. They've maybe tried a search. But running four AI powered research workflows that feed directly into their
[08:25]
business growth. building a system where deep research is doing competitor analysis, lead research, and content strategy automatically. Most haven't done it. That window won't stay open. As these tools get more powerful and more
[08:39]
coverage builds around them, the early movers will have a significant head start. Inside the AI profit boardroom, we've already built out the daily tutorials walking you through exactly how to use perplexity computer and deep
[08:53]
research for real business tasks. competitor research, lead qualification, market analysis, content planning, four coaching calls every week where you bring your actual setup, ask live questions, and get walked through
[09:08]
exactly what to do next. A prompt library with tested perplexity prompts for lead genen research workflows and automation. and a member map so you can connect with other members in your area who are already running this same stack
[09:24]
and can give you realworld input. This is the fastest way to actually build it. Link in the description or go to aiprofitboardroom.com. And if you want to start for free, come join the AI success lab. It's our free
[09:39]
community and right now there are over 92,000 members in there learning AI automation together. That's where all the video notes from this episode live. Links to every tool, every workflow, every prompt I mentioned today. You'll
[09:54]
every prompt I mentioned today. You'll get access to over 100 AI use cases, including the full Plexity research workflows. The community is active around the clock. There's always someone online, always someone who's already
[10:07]
tried the thing you're stuck on, and always new tools being shared. It's completely free to join. No catch. All the links are in the comments and the the links are in the comments and the description below.